What problem does it solve? After a module implementation lands, newly stabilized business terms and user roles often stay buried in module specs instead of being registered in the root PRODUCT.md, causing inconsistent naming across modules. This Skill reconciles those structured declarations against the existing product glossary so cross-module reuse stays consistent. ## Core Features & Use Cases - Structured Term Extraction: Reads only explicit declarations from spec.md glossary tables, user role tables, D-numbered decisions, and accepted_deltas affects entries—never guesses from prose, quotes, or code. - Whitelist and Skip Filtering: Applies a built-in whitelist of common tech and business words plus an optional .term-skip.json to suppress false positives. - Doc-Impact Integration: Emits pending kind=term items targeting PRODUCT.md through doc-impact.py, returning JSON with new_terms, new_roles, skipped, whitelisted, and registered lists. - Use Case: After a settlement module lands, the detector finds "结算批次" and "平台审核员" declared in the module spec but missing from PRODUCT.md, and queues them as documentation impact items for the compile step. ## Quick Start Run the term detector after a module lands to find any new business terms or roles that should be registered in PRODUCT.md.
